INDEX

Explanations

patterns and sequences related to coding and data structures

oai_token-act-pair · gpt-4o-mini Triggered by @bot

Abbreviations ending in "s"

np_acts-logits-general · gemini-2.0-flash

abbreviations and acronyms

np_acts-logits-general · gemini-2.5-flash-lite

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B-IT @ 20-gemmascope-res-16k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_16k/average_l0_91

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.20.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 pleaſure

-0.73

 Majefty

-0.73

 fubject

-0.52

SocketChannel

-0.51

wouldn

-0.48

Efq

-0.47

 ſever

-0.47

 purpoſe

-0.47

 myſelf

-0.47

 étag

-0.46

POSITIVE LOGITS

1.13

es

0.94

bs

0.92

ses

0.88

rs

0.87

ens

0.85

ks

0.85

zs

0.84

os

0.83

ras

0.83

Activations Density 1.551%

patterns and sequences related to coding and data structures

Abbreviations ending in "s"

abbreviations and acronyms

No Comments

No Known Activations